source: trunk/scripts/migrate-authors-to-task-tags @ 1184

#! /usr/bin/env php
<?php

require_once(dirname($argv[0]) . "/utilities.php");
require_once(IA_ROOT_DIR . "common/db/task.php");
require_once(IA_ROOT_DIR . "common/tags.php");
db_connect();

$tasks = task_get_all();
$author_tags = array();
foreach($tasks as &$task) {
    // Special cases to make my life easier.
    if ($task["author"] == "") {
        $task["author"] = "Necunoscut";
        task_update($task);
    }
    if ($task["author"] == "Alin Burtza") {
        $task["author"] = "Alin Burta";
        task_update($task);
    }
    if ($task["author"] == "Maria si Adrian Nita") {
        $task["author"] = "Maria Nita si Adrian Nita";
        task_update($task);
    }

    // Split authors separated by , & ; or si
    $authors = preg_split("/,|&|;|(\ssi\s)/", $task["author"]);
    $authors = array_map('trim', $authors);
    $author_tags = array_unique(array_merge($author_tags, $authors));
}

// Split every author into individual words so that when compared "A B"
// is the same as "B A".
$authors = array();
foreach($author_tags as $author) {
    $authors[$author] = preg_split("/[^a-zA-Z.]/", $author);
    sort($authors[$author]);
}

// Check if one array is completely included in another.
// Used to group "Cosmin-Silvestru Negruseri" and "Cosmin Negruseri" together.
function included($a, $b) {
    $intersection = array_values(array_intersect($a, $b));
    sort($intersection);
    return $intersection == $a;
}

// Group same authors together.
$grouped = array();
foreach($authors as $key => $author) {
    if (isset($grouped[$key])) {
        continue;
    }
    // Skip if included in other.
    $ok = true;
    foreach($authors as $other) {
        if ($author == $other) {
            continue;
        }
        if (included($author, $other)) {
            $ok = false;
            break;
        }
    }
    if (!$ok) {
        continue;
    }

    $group = array();
    foreach($authors as $key_other => $other) {
        if (included($other, $author)) {
            $group[] = $key_other;
        }
    }

    // Choose a tag for the group
    // The admin should check that each tag is in First_Name Last_Name form.
    $tag = read_line(sprintf(
        "Alege un tag de forma 'Prenume Nume' pentru grupa:\n- %s\n",
        implode("\n- ", $group)), $group[0]);
    foreach($group as $author) {
        $grouped[$author] = $tag;
    }
}

// Add author tags to tasks
foreach($tasks as $task) {
    // Split authors separated by , & or si
    $authors = preg_split("/,|&|;|(\ssi\s)/", $task["author"]);
    $authors = array_map('trim', $authors);

    tag_clear("task", $task["id"], "author");
    foreach($authors as $author) {
        tag_add("task", $task["id"], tag_assign_id(array(
            "name" => $grouped[$author],
            "type" => "author",
            "parent" => 0
        )));
    }

    // Change all authors in ia_task to match tags until we remove the column.
    $new_authors = tag_build_list("task", $task["id"], "author");
    if ($task["author"] != $new_authors) {
        log_print(sprintf("Changing task '%s' author from '%s' to '%s'",
            $task["id"], $task["author"], $new_authors));
        $task["author"] = $new_authors;
        task_update($task);
    }
}

db_query("ALTER TABLE ia_task DROP COLUMN author");

?>